A Heuristic Search for Optimal ParameterValues of Three Biokinetic Growth Models for Describing Batch Cultivations of Streptococcus Pneumoniae in Bioreactors

摘要

Simulated annealing (SA) is a stochastic search procedure which can lead to a reliable optimization method. This work describes a dynamic mathematical model for Streptococcus pneumoniae batch cultivations containing 8 unknown parameters, which were calibrated by a SA algorithm through the minimization of an evaluation function based on the performance of the model on real experimental data. Three kinetic expressions, the Monod, Moser and Tessier equations, commonly employed to describe microbial growth were tested in the model simulations. SA convergence was achieved after 13810 interactions (about 10 minutes of computing time) and the Tessier equation was identified as the kinetic expression which provided the best fit to the cultivation dataset used for parameter estimation. The model comprising the Tessier equation, estimated parameter values supplied by SA and mass balance equations was further validated by comparing the simulated results to 3 experimental datasets from new cultivations carried out in similar conditions.